Heart Disease Prediction using Machine Learning Algorithm

International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 5 Issue 2, January-February 2021 Available Online: www.ijtsrd.com e-ISSN: 2456 - 6470

Heart Disease Prediction using Machine Learning Algorithm

Ravi Kumar Singh, Dr. A Rengarajan

Department of Master of Computer Applications, Jain Deemed to be University, Bengaluru, Karnataka, India

ABSTRACT

How to cite this paper: Ravi Kumar Singh

Nowadays, Heart disease has become dangerous to a human being, it effects | Dr. A Rengarajan "Heart Disease

very badly to human body. If anyone is suffering from heart disease, then it
leads to blood clotting. Heart disease prediction is very difficult task to predict
in the field of medical science. Affiliation has predicted that 12 million people in
fail horrendously every year as a result of heart disease. In this paper, we
propose a k-Nearest Neighbors Algorithm (KNN) way to deal with improve the
exactness of heart determination. We show that k-Nearest Neighbors and
Algorithm (KNN) have better accuracy than random forest algorithm for
viewing heart disease. The k-Nearest Neighbors Algorithm give more precise
and exact outcome. We have taken 13 attributes in the dataset and a target
attribute, by applying machine learning we achieved 84% accuracy in the

heart disease detection.

KEYWORDS: Machine Learning, k-Nearest Neighbors classifier, Decision Tree

classifier, Random Forest Classifier, Jupyter

INTRODUCTION

Heart disease forecast is one of the most notable pointin the
machine learning field for expectation. It clusters the blood
to all aspects of the body. If the blood not siphons to every
part of the body, at that point the brain and different organ
will stop work and the person may die. It is hard to recognize
heart disease on account of few factors, for example,
diabetes, hypertension, high cholesterol, heart beat rate and
various other factors. As per World Health Organization
heart related disease are liable for taking 17.7 million lives
every year, 31% of all over worldwide. In India, heart
disease has become the main source of mortality. Heart
disease has killed 1.7 million Indian in 2016, as indicated by
the 2016 worldwide weight of infection report.

In clinical science coronary illness is one of the huge
challenges, because a lot of parameter and technicality is
involved for predicting this disease. Machine learning could
be a superior decision for accomplishing high precision for
heart disease as well as another disease and its diverse
information types under different condition for predicting
the heart disease calculation, for example, Naive Bayes,
Decision Tree, KNN, Neural Network are utilized to predict
risk of heart algorithm and its speciality such as Naive Bayes
is utilized for predicating heart disease, while Decision Tree
is utilized to give ordered report to the heart disease, though
the Neural Network give chances to limit the mistake for
predication of heart disease. All these procedures are
utilized in old patient record for getting expectation about
new patient. The expectation for heart disease encourages
doctor to predict heart disease in early stage so that he can
save millions of lives.

Prediction using Machine Learning
Algorithm" Published || = }
International
Journal of Trend in
Scientific Research
Development
(ijtsrd), ISSN: 2456-
6470, Volume-5 |
Issue-2, February |. :
2021, pp.183-187, URL:
www.ijtsrd.com/papers /ijtsrd38358.pdf

Copyright © 2021 by author(s) and
International Journal of Trend in Scientific
Research and Development Journal. This
is an Open Access article distributed
under the terms of

the Creative
Commons Attribution . i
License (CC BY 4.0)

(http://creativecommons.org/licenses/by/4.0)

This overview paper is committed for a review in the field of
machine learning technique in heart disease. Later aspects of
the overview paper will discuss about different machine
learning calculation for heart disease and their comparison
on different parameters. It also shows future outline of
machine learning calculation in heart disease. This paper
gives a profound analysis in the field of predicting heart
disease.

RELATED WORKS

Heart is one of the main organ of human body, it plays vital
function of blood siphoning in human body which is as
fundamental as the oxygen of human body so there is
consistently need of insurance of it, this is one of the main
explanation behind the analysts to work on it. So there are
number of specialists dealing with it. There is consistently
need of examination of heart related things either analysis or
expectation or you can say that assurance of heart disease.
There are different fields like artificial intelligence, machine
learning, data mining that contributed on this work. Here, we
will discuss some of them.

Some of the analysts have taken a shot of information about
the expectation of heart disease. Kaur et al. have worked on
this and characterize how the interesting pattern and
information are gotten from a huge dataset. They perform
exactness correlation on different machine learning and
information mining 453 methodologies for discovering
which one is best among at that point and get the outcome
on the kindness of SVM.

@IJTSRD | Unique Paper ID - JTSRD38358 _ |

Volume-5|Issue-2 |

January-February 2021 Page 183

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470

Zhao et al. (2017) built up a framework for heart disease
classification by utilizing two datasets, one from Shanghai
Shuguang Hospital and another in UCI coronary disease
dataset. The model uses support Vector Machine calculation
alongside PCA, CCA and DMPCCA which are utilized for
include extraction and combination. The _ general
investigation come about that DMPCCA gave the best
outcome.

Ganesan et al. (2019) utilize IOT innovation for expectation
and conclusion of heart disease by taking UCI dataset and
applied J48 classifier, Logistic Regression, Multiplayer
Perception, and SVM utilizing Java on Amazon cloud. In this
examination J48 gives 91.48%, SVM gave 84.07%, LR gave

Literature survey
Authors

Year

Palaniappan
and Awang

83.07%, and MPL gave 78.14% exactness and inferred that
J48 beats every other calculation.

PROJECT SCOPE AND OBJECTIVES

The primary goal of this examination is to develop a heart
forecast framework. The system can find information related
with heart disease from the historical heart data set to
implement the classifier that classifies the disease according
to the contribution of the client and reduce the cost of the
medical test. The scope of the project is to execute machine
learning calculation to bigger dataset helps to improve the
accuracy of results. Utilizing of machine learning procedure
gives more exact outcomes than more experienced doctor.
By this clinical choice with computer-based patient record
could decrease medical error and improve patient result.

Description

The authors proposed to develop a model Intelligent Heart Disease Prediction
2008 | System (IHDPS) utilizing information mining procedures to be specific Naive
Bayes, Decision Tree, and Neural Network.

The authors proposed that neural network was best survey in information
2 Bhatla and Jyoti 2012 o- prop - é f
mining methods to anticipate heart disease.

The creators proposed three mainstream information mining calculation CART

Chaurasia and Pal 2013

(Classification and Regression Tree), ID3 (Iterative Dichotomized 3) and

Decision Table (DT) separated from a choice tree to foresee heart disease.
The authors proposed to utilize diverse characterization procedures in coronary

Boshra Brahmi etal. | 2015

illness determination like J48 Decision Tree, K-Nearest Neighbors (KNN), Naive

Bayes (NB) and SMO to classify dataset.

c . Vembandasamy et 5015

The authors proposed Naive Bayes algorithm in data mining technique which
serves diagnosis of heart disease patient.

6 Can aeral. 5016 The authors propose an efficient mechanism to predict heart disease by mining
the data from health record.

7 K. Gomathi et al. 2016

The authors proposed to analysis information mining methods to foresee various
kinds of sicknesses like heart disease, diabetes and bosom disease and So on.

The authors proposed of this examination is to dissect directed AI calculation to
Ayon Dey etal. 2016 >.
anticipate heart disease.

Requirement Analysis

Tools

Anaconda

Anaconda is an open-source appropriation for python andR
programming language. It is utilized for information science,
machine learning, profound learning, and so on. With the
availability of more than 300 libraries for information
Science, it turns out to be genuinely ideal for any developer
to work on anaconda for information science. Anaconda
helps in improved bundle the board and sending. Anaconda
accompanies the wide assortment variety of tools to
effectively gather information from different source using
different machine learning and machine learning
calculations. It is developed and maintained by
Anaconda.inc., which was developed by Peter Wang and
Travis Oilphant in 2012.

Hardware Requirements

> Operating System: Windows 10

> Processor: Intel(R)Pentium(R) CPU N3710 @1.60GHz
1.60GHz

> System Type: 64-bit operating system, x64-based
processor

> Installed Ram: 4.00 GB

Software Requirements

Jupyter Notebook

The Jupyter Notebook is an open-source web application
that permits you to make and offer chronicles that contain
live code, condition, perceptions and story text. Utilization

include: information cleaning and_ transformation,
mathematical simulation, measurable displaying,
information representation, machine learning, and

significantly more.

Python

Python is a universally useful deciphered, intelligent, object-
arranged and elevated level programming language. It was
developed by Guido van Rossum during 1985-1990. Like
Perl, python source code is additionally accessible under the
GNU General Public License (GNL). Its Error! Bookmark not
defined. and object-oriented approach aim to
help programmers write clear, logical code for small and
large-scale projects.

Python Libraries
> Numpy

> Pandas

> Matplotlib

> Sklearn

@IJTSRD | Unique Paper ID - JTSRD38358 _ |

Volume-5|Issue-2 |

January-February 2021 Page 184

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470

Material and Methods

Dataset Used for Research

The dataset consists of 303 individual data. There are 14
columns in the dataset, which are described below.

1. Age: displays the age of the individual.

2. Sex: displays the gender of the individual using the
following format:
1 = male
0 = female

3. Chest Pain type: shows the kind of chest-torment
experienced by the individual utilizing the accompanying
organization t:

1 = typical angina

2 = atypical angina

3 = non — anginal pain
4 =asymptotic

4. Resting Blood Pressure: shows the resting pulse
estimation of a person in mmHg (unit)

5. Serum Cholestrol: shows the serum cholesterol in mg/dl
(unit)

6. Fasting Blood Sugar: looks at the fasting glucose
estimation of a person with 120mg/dl. In the event that
fasting glucose > 120mg/dl at that point: 1 (valid)

7. Resting ECG: displays resting electrocardiographic
results
0 = normal
1 = having ST-T wave abnormality
2 =left ventricular hyperthrophy

8. Max heart rate achieved: displays the max heart rate
achieved by an individual.

9. Exercise induced angina:
1 = yes
O =no

10. ST depression induced by exercise relative to rest:
displays the value which is an integer or float.

11. Peak exercise ST segment:
1 = upsloping
2 = flat
3 =downsloping

12. Number of major vessels (0-3) colored by flourosopy:
displays the value as integer or float.

13. Thal: displays the thalassemia:

3 = normal
6 = fixed defect
7 =reversible defect

14. Diagnosis of heart disease: Displays whether the
individual is suffering from heart disease or not:
0 = absence
1 = present.

Classification Techniques

Procedures In AI and measurements, grouping is a directed
learning approach in which the PC program gains from the
information and afterward utilizes this figuring out how to

order groundbreaking perceptions. At the end of the day, the
preparation dataset is utilized to acquire better limit
conditions which can be utilized to decide each target class;
when such limit conditions are resolved, next undertaking is
to foresee the objective class

Machine learning is a field of study and is concerned with
algorithms that learn from examples. There are many
different types of classification tasks that you may encounter
in machine learning and specialized approaches to modelling
that may be used for each.

K-Nearest Neighbor Algorithm (KNN)

K nearest neighbors is one of the easiest machine learning
calculation is dependent on supervised learning procedure.
K-NN calculation accepts the closeness between the new
case and available cases and put the new case into the
classification that is generally like the accessible
classification. K-NN calculation can be utilized for regression
just as for classification issue. K-NN is a non-parametric
calculation, which implies it doesn’t make any presumption
on hidden information.

Pros:

> Basic Algorithm and consequently simple to decipher the
forecast. Quick calculation time.

> Used for both classification and regression.

Cons:

> Does not work well for large dataset.

> Prediction is very costly.

> Pooratclassifying data points ina boundary where they
can be classified one way or another.

Random Forest Classifier

Random Forest is one of the most prestigious and most
remarkable machine learning calculations. It is one sort of
machine learning calculation that is called Bagging or
Bootstrap Aggregation. So, as the access an incentive from an
information test, for example, mean, the bootstrap is very
powerful statistical approach. Here, lots of information are
taken, the mean is determined, after that all the mean value
are averaged to give a superior expectation of the mean value.
In bagging, a similar strategy is utilized, but instead of
estimating the mean of each information test, decision tree is
commonly utilized.

Advantage of Random Forest:

> Random Forest Algorithm is exact outfit learning
calculation.

> Random Forest runs efficiently for large scale data sets.

> Itcan handle hundreds of input variables.

Disadvantage of Random Forest:

> Features need to have some predictive power else they
won't work.

> Forecasts of the trees should be uncorrelated.

> Appears as black box.

Decision Tree Classifier

Decision Tree Classifier is a basic and generally utilized
grouping procedure. It applies a waterway forward plan to
take care of the grouping issue.. Decision tree classifier
represents a progression of deliberately made inquiries
concerning the characteristics of the test record. Decision

@IJTSRD | Unique Paper ID - JTSRD38358 _ |

Volume-5|Issue-2 |

January-February 2021 Page 185

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470

Trees (DTs) area non-parametric directed learning method
used for classification and regression. It is a Supervised
Machine Learning where the information is constantly part
as indicated by a specific boundary.

Decision Tree consists of:

> Nodes: Test for the estimation of a specific quality.

>» Edges/ Branch: Compare to the result of a test and
associate with the following hub or leaf.

> Leaf nodes: Terminal hubs that anticipate the result
(speak to class marks or class appropriation).

Experiment

The Proposed Method

Heart disease is the main source of death among all the
diseases, even cancer. The quality of people facing heart
disease is on a raise every year. The prompts for its initial
finding and treatment. Because of absence of source in the
medical field, the prediction of heart disease might be a
issue. Use of suitable technology can be useful to the medical
society and patient. The issue can be settled by embracing
machine learning techniques. In my project, I would be
taking a shot at basic machine learning classification model.
And using this model I could prepare my model utilizing the
information which comprise of different attribute like age,
sex, cp, blood pressure, skin thickness and so on and based

Result and Discussion
Correlation Matrix

on this attribute I would anticipate the outcome for a patient
whether he is experiencing heart disease or not. This paper
has Random Forest classifier, KNN (K-Nearest neighbour
classifier) & Decision Tree classifier - three techniques for
the effective prediction of heart disease. It analyses the
efficiency & accuracy of the three techniques to choose them
the best.

The figure below shows the number of the heart disease
cases.

160
140
170

100

count

6 8 6

co

target

0 = absence 1 = present

Let's see the correlation matrix of features. From this graph, we can observe that some features are highly correlated and some

are not.

z

“1

8

g

é

E

3

a

5

“ 0.085 0028 014 #1042 0.35

5

age sex @ trestbps chol fos restecg thalach exang oldpeak slope c thal target
This figure shows the correlation matrix

@IJTSRD | Unique Paper ID-IJTSRD38358 | Volume-5|Issue-2 | January-February 2021 Page 186

International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470

K-Nearest Neighbors Classifier:

K Nearest Neighbors is anon-parametric strategy utilized for
grouping. It is lazy learning figuring where all computation is
surrendered until gathering. Itis otherwise called case based
learning calculation, where the capacity is approximated
locally. This algorithm is used when the amount of data is
large and there are non-linear decision boundaries between
classes. KNN explains a categorical value using the majority
votes of nearest neighbors. Not only for classification, KNN
can be used for function approximation problem.

K Neighbors Classmier scores for different k values

a, OT
a6 4 4, 6.86) 8, Of)
pf \ 10, O35)
84 | a, “Ni a haf Fb)
| (5, 0.63} 12. 6 O.A7

¢ O82 |
: 16, O81)

oa0 4 sm, o8

7, 0.78)

[po a ee = = —==

f1, 0.76)

f 1 7 ' i is f t T T t t t t t ee
P2 Fa SETAE SIMWUPLEMSEBEU BY Dw
Humber of Neighbors (€)

This figure shows the K Neighbors Classifier scores

Random Forest Classifier:

Random forest is a regulated learning calculation. It very
well may be utilized for order and relapse. It is
straightforward and simple to execute. A backwoods is
contained trees. This classifier makes choice trees on
haphazardly chose information tests, gets forecast from each
tree and chooses the best arrangement by methods of
casting a ballot. The random forest composed of multiple
decision trees. It creates a forest of trees.

Ruancdiam Forest! Classes voor fo7 Seren! nuoembes: of ef rralor
ia

Ae yma | Ps =i i EF
ier of eae

This figure shows the Random Forest Classifier scores.

Decision Tree Classifier

This classifier falls under the category of supervised
learning. It very well may be utilized to take care of relapse
and characterization issues. We can utilize this calculation
for issues where we have ceaseless yet in addition
unmitigated info and target highlights. It is the best machine
learning calculation utilized for depicting the tree in a
graphical way.

Decioon Tree Classifier scores for different number of maximum features

ores
La

115, 0.64)
Lid J 45° 6.7 BORE el SA a
Max leatures

This figure shows the Decision Tree Classifier scores

Conclusion

Machine Learning plays an important role in various fields
such as Healthcare, Stocks & Marketing, Banking, Weather
Forecast and so on. With the help of KNN Algorithms it
become easy to evaluate and fetch meaningful information
from them. In KNN by using the various K- values of the K-
NN classifier the accuracy of the model increases
simultaneously, this study aims to accurately predicting
whether a given patient is suffering from diabetes or not.
Finally the accuracy of my model comes close to 84 % and
for any new patient it could easily predict whether the
patient is having diabetes or not.

Bibliography

[1] S. E.-S. S. I. D. K. A. A. F Ali, "A smart healthcare
monitoring system for heart disease prediction based
on ensemble deep learning and feature fusion," 2020.

[2] C.T.G.S.S Mohan, "Effective heart disease prediction
using hybrid machine learning techniques," 2019.

[3] M.R.M.I.M.1I.S Nashif, "Heart disease detection by
using machine learning algorithms and a real-time
cardiovascular health monitoring system," 2018.

[4] Y.H.K.H.L.W.L.W.M Chen, "Disease prediction by
machine learning over big data from healthcare
communities," 2017.

[5] A.A. AA Soofi, “Classification techniques in machine
learning: applications and issues," 2017.

[6] S.S.K Deepika, "Predictive analytics to prevent and
control chronic diseases," 2016.

[7] D.P.K Gomathi, "Multi Disease Prediction using Data
Mining Techniques, 2016.

[8] J. S. N.S. A Dey, “Analysis of supervised machine
learning algorithms for heart disease prediction with
reduced number of attributes using principal
component analysis, 2016.

[9] M.S. B Bahrami, "Prediction and Diagnosis of Heart
Disease by Data Mining Techniques,’ 2015.

[10] R.S.E.D.K Vembandasamy, "Heart diseases detection
using Naive Bayes algorithm," 2015.

[11] E. A. Y. K. AF Otoom, "Effective diagnosis and
monitoring of heart disease," 2015.

[12] S.P.V Chaurasia, “Early prediction of heart diseases
using data mining techniques," 2013.

[13] S. S. G Parthiban, “Applying machine learning
methods in diagnosing heart disease for diabetic
patients," 2012.

[14] K.J. N Bhatla, "An analysis of heart disease prediction
using different data mining techniques," 2012.

[15] R. A. S Palaniappan, “Intelligent heart disease
prediction system using data mining techniques,"
2008.

@IJTSRD | Unique Paper ID - JTSRD38358 _ |

Volume-5|Issue-2 |

January-February 2021 Page 187